British Journal of Clinical Pharmacology. Daniel Atwine 1,2,4, Maryline Bonnet 1,3,4 and Anne-Marie Taburet 5,6

Transcription

1 British Journal of Clinical Pharmacology Br J Clin Pharmacol (2018) SYSTEMATIC REVIEW AND META-ANALYSIS Pharmacokinetics of efavirenz in patients on antituberculosis treatment in high human immunodeficiency virus and tuberculosis burden countries: A systematic review Correspondence Daniel Atwine, Epicentre Mbarara Research Centre, P.O Box 1956, Mbarara, Uganda. Tel: ; daniel.atwine@epicentre.msf.org; daniel.atwine@must.ac.ug Received 27 November 2017; Revised 7 March 2018; Accepted 26 March 2018 Daniel Atwine 1,2,4, Maryline Bonnet 1,3,4 and Anne-Marie Taburet 5,6 1 Epicentre Mbarara Research Centre, Mbarara, Uganda, 2 Mbarara University of Science and Technology, Mbarara, Uganda, 3 IRD UMI233 TransVIHMI-INSERM U1175, Montpellier, France, 4 University of Montpellier, Montpellier, France, 5 Bicetre hospital, Paris, France, and 6 UMR 1184, INSERM,CEA,UniversitéParis-Sud Keywords pharmacokinetics, genetics and pharmacogenetics, HIV/AIDS < infectious diseases, antiretrovirals < infectious diseases, adverse drug reactions AIMS Efavirenz (EFV) and rifampicin isoniazid (RH) are cornerstone drugs in human immunodeficiency virus (HIV) tuberculosis (TB) coinfection treatment but with complex drug interactions, efficacy and safety challenges. We reviewed recent data on EFV and RH interaction in TB/HIV high-burden countries. METHODS We conducted a systematic review of studies conducted in the high TB/HIV-burden countries between 1990 and 2016 on EFV pharmacokinetics during RH coadministration in coinfected patients. Two reviewers conducted article screening and data collection. RESULTS Of 119 records retrieved, 22 were included (two conducted in children), reporting either EFV mid-dose or pre-dose concentrations. In 19 studies, median or mean concentrations of RH range between 1000 and 4000 ng ml 1, the so-called therapeutic range. The proportion of patients with subtherapeutic concentration of RH ranged between 3.1 and 72.2%, in 12 studies including one conducted in children. The proportion of patients with supratherapeutic concentration ranged from 19.6 to 48.0% in six adult studies and one child study. Five of eight studies reported virological suppression >80%. The association between any grade hepatic and central nervous system adverse effects with EFV/RH interaction was demonstrated in two and three studies, respectively. The frequency of the CYP2B6 516G > T polymorphism ranged from 10 to 28% and was associated with higher plasma EFV concentrations, irrespective of ethnicity. CONCLUSIONS Anti-TB drug coadministration minimally affect the EFV exposure, efficacy and safety among TB-HIV coinfected African and Asian patients. This supports the current 600 mg EFV dosing when coadministered with anti-tb drugs The British Pharmacological Society DOI: /bcp.13600

2 D. Atwine et al. Introduction Human immunodeficiency virus (HIV) infection is still a global public health concern, especially in Africa and Asia [1]. Sub-Saharan Africa remains most severely affected, accounting for nearly 70% of the people living with HIV worldwide [1]. The estimated risk of developing tuberculosis (TB) in people living with HIV ranges between 26 and 31 times greater than in those without HIV infection [2]. The highest TB incidence rates among HIV patients are reported in Africa and Asia. The overall TB mortality rate among HIV patients is about 6 times higher in Africa (30 per population) than the global average (5.3 per population), with approximately 75% of all deaths occurring in Sub-Saharan Africa [3]. The recommended first-line antiretroviral treatment for adults and adolescents consists of two nucleoside analogues, reverse-transcriptase inhibitors plus a non-nucleoside reversetranscriptase inhibitor (NNRTI) or an integrase inhibitor (INSTI) [4]. The recommended World Health Organization reverse-transcriptase inhibitor backbone was zidovudine + lamivudine (3TC) and recently tenofovir disoproxyl fumarate + 3TC. Tenofovir disoproxyl fumarate and 3TC are eliminated unchanged through the kidney, and are substrates of uptake and efflux transporters. Consequently, they are less likely to exhibit potent drug drug interactions with anti-tb treatment than drugs whose elimination involves drug metabolizing enzymes such as cytochrome P-450 (CYP) [4]. The recommended NNRTI is efavirenz (EFV) given its proven high virological efficacy, its availability as a fixed-dose combination administered at 600 mg dose once-daily, under generic formulation and preferably in the night to minimize central nervous system (CNS) adverse events, and to ensure good adherence. EFV is well incorporated within the national guidelines of low income and high HIV burden countries [5 7]. There is reassuring data regarding its safety in pregnancy [8] and improved efficacy compared to the former widely used NNRTI, nevirapine [9]. Although dolutegravir, an HIV-integrase inhibitor, has been found more tolerable compared to EFV, and with increasing availability in resource limited countries, limited experience of its use in HIV-TB coinfected patients will make EFV-based antiretroviral therapy (ART) regimen the cornerstone of HIV treatment in these patients for many years [4, 10]. EFV is mostly metabolized by CYP2B6 through hydroxylation to inactive metabolite 8-hydroxy EFV, and to a lesser extent by CYP2A6 into 7-hydroxy EFV [11, 12], with CYP3A4/5 and CYP1A2 playing a minor role in this step [12, 13]. The main metabolite, 8- hydroxy EFV is further hydroxylated primarily by CYP2B6 to form 8, 14-hydroxy EFV. The oxidative metabolites undergo conjugation by UDP-glucuronyltransferase (UGT2B7) pathway [14] and are excreted in the urine as glucuronides [15]. There are reports thatcyp2a6andugt2b7onlyplayasignificant role in the efavirenz pathway when CYP2B6 activity is impaired [16]. EFV plasma concentrations below 1000 ng ml 1 in samples collected 8 20 h post intake, have been associated with increased risk of virological failure in HIV-infected patients, while concentrations above 4000 ng ml 1 have been associated with risk of CNS adverse effects [17, 18]. There is a wide interindividual variability in EFV concentrations [18] that is partially explained by genetic factors as shown by the strong association between CYP2B6 516G > T single nucleotide polymorphism and EFV exposure [19]. The CYP2B6 516G > T is a common polymorphism that has been consistently associated with reduced enzyme activity, higher EFV exposure and increased toxicity [11, 19, 20]. By contrast, there is no clear evidence supporting the association with gender and body weight [15, 21]. With regard to drug susceptible TB, a 6-month regimen is broken down into an intensive 2-month phase involving isoniazid (H, 5 mg kg 1 ), rifampicin (R, 10 mg kg 1 ), pyrazinamide (Z, 25 mg kg 1 ) and ethambutol (15 mg kg 1 ) followed by 4 months of continuation phase with R and H [22]. Both R and H are the cornerstone drugs within this regimen [23, 24] that has a very good efficacy [25]. This regimen is used with fixed-dose combination and administered once daily [24]. R has a strong bactericidal activity [25, 26] given its ability to inhibit transcription by binding with high affinity to bacterial DNA-dependent RNA polymerase [27 29] and the best sterilizing drug to prevent relapses of TB. R is also a potent inducer of several liver or gut drug metabolizing enzymes, especially isoenzymes of CYP, mainly isoenzyme CYP3A4 and CYP2B6. This results in enhanced NNRTI drug metabolism and may lead to subtherapeutic NNRTI plasma concentration during coadministration [30]. In healthy volunteers, EFV area under the curve, maximum concentration and minimum concentration (C min ) are reduced by 26, 20 and 32% when coadministered with R as compared to EFV alone, respectively, which led to the Food and Drug Administration recommendation of an increase in EFV dosing to 800 mg once a day when combined with TB drugs [30, 31]. However, due to the potential of increased risk of CNS toxicity with the increase of EFV dose and reassuring virological response in coinfected patients receiving EFV at 600 mg once daily in high HIV burden countries, it is recommended to maintain EFV at usual dose (600 mg day 1 once daily) [17, 32 36]. The other cornerstone anti-tb drug, H, is metabolized mainly through N-acetyltransferase type 2 (NAT2) and was demonstrated in vitro to have an inhibitory effect on several cytochrome P450 enzymes (CYP2C19, CYP1A2, CYP2A6, CYP2C19 and CYP3A4) [37], and especially through its effect on CYP2A6 metabolic pathway, could impact the relationship between combined R and H (RH) and EFV, rendering the inducing effect of RH less potent than R alone [38, 39]. Such effect could be different according to patient s CYP2B6 516 G > T genetic polymorphism [40]. Although the goal of providing ART has over time expanded from saving lives to include long-term virus control and to reduce transmission [41], but the effects of ART coadministration with other treatments that risk impairment of the ART blood concentrations may impend the attainment of this goal. A strong evidence base to support such public health approaches is needed to ensure good results from delivery of treatment at scale without compromising quality [41] but also safety and efficacy. Although attempts have been made in describing the pharmacokinetics (PK) and pharmacogenetics of EFV with RH coadministration [15, 42 44], an extensive focus on the world s highest 1642 Br J Clin Pharmacol (2018)

3 Pharmacokinetics of efavirenz with antituberculosis treatment HIV/TB burden countries that may be affected most by drug drug interactions is lacking. We conducted a systematic review to gather existing information on the PK of EFV during RH coadministration among TB and HIV high-burden countries. We assessed the effect of body weight, sex, EFV dosing and the CYP2B6 homozygous slow metabolizer genetic polymorphism, CYP2B6 516 G > T, on the EFV concentrations during RH coadministration, and the effect of the EFV PK results on the virological response, CNS and hepatic toxicity. Methods Study eligibility criteria A study was considered eligible for inclusion if it was a randomized controlled trial, cohort, case control or crosssectional study, that report PK parameters of EFV (minimum or mid-dose concentrations at least) following coadministration with RH in TB/HIV coinfected patients for at least 4 weeks (that is to ensure a minimum steady state) and conducted in one of the World Health Organization TB/HIV high burden countries [45]. Studies that enrolled patients with comorbidities that require coadministration of other drugs with known interaction with EFV and studies enrolling patients on anti- TB prophylaxis or using other rifamycins such as rifapentine and rifabutin were excluded. Only studies published in English were included. Search strategy We conducted this review according to PRISMA guidelines [46]. We identified relevant articles through a systematic search of Cochrane Library, EMBASE.COM and MEDLINE (via OvidSP) published from 1 January 1990 to 31 August 2016 in the English language. The choice of 1990 as start point was based on the consideration that in most of the TB/HIV high-burden countries, access to antiretroviral treatment took place after 1990, with EFV receiving Food and Drug Administration approval in 1998 [47]. We also searched the Web of Science and carried out manual searches (hand searching) to retrieve other reports of studies that are reported in journals, conference proceedings, bibliographies of review articles and retrieved articles, monographs, and sources other than those mentioned above. We used the following abbreviated search strategy: ( Efavirenz or Stocrin or Sustiva ) and ( Rifampicin or RIFAFOUR or RIFAMPIN or RIFAMYCIN ) and( pharmacokinetics or drug assay or plasma drug concentration or Ctrough or pharmacology or drug interaction or drug drug interaction or Efavirenz concentration or Rifampicin concentration or non-nucleoside reverse transcriptase inhibitors concentration ). Bibliography search and screening of titles and abstracts were done by one reviewer (D.A.), duplicate records were eliminated and full texts of potentially relevant articles retrieved. The selection was validated by a second reviewer (M.B.), blinded to the initial assessment. Full texts retained through this process were independently screened by two reviewers (D.A. and A.M.T.). Disagreements were examined by the third reviewer (M.B.). Records with inaccessible full text but with author contact details were retrieved after contacting authors by . Abstract only records were excluded from further data collection processes. Data collection and analysis Data collected from each study were recorded by one reviewer (D.A.) in a standardized data extraction form (see Appendix S1) and validated by the second reviewer (A.M.T.). Authors were contacted for clarification whenever needed. All discrepancies were discussed and resolved by consensus between the three reviewers (D.A., A.M.T. and M.B.). Data extraction forms were entered into a database using Epi Info software (V7.2, Atlanta, GA USA) and analysis used the Stata software (v. 13, College Station, TX, USA). We performed descriptive presentation of the studies and patients characteristics. EFV mid-dose concentrations measured 12 h postdose (C 12 )orpredoseconcentration (C min ) measured before next dose intake were the PK parameters chosen as a surrogate of EFV exposure based on their availability in all selected articles. They were presented with or without RH coadministration globally, by sex, body weight and EFV dose. Proportions of patients with subtherapeutic (<1000 ng ml 1 ) or supratherapeutic EFV concentrations (>4000 ng ml 1 ) [18] were presented graphically per study and geographical region. EFV concentrations in homozygous slow metabolizer patients carrying CYP2B6 516TT gene and extensive metabolizers carrying CYP2B6 516GG gene [48], were presented graphically by study. Results for heterozygous patients (CYP2B6 516GT) were not shown. Results Of the total 119 records retrieved, 22 records were included in the analysis (Figure 1). Of these, 20 studies had data on at least mean or median EFV C 12 (n = 16) or predose C min (n = 6) measured either in the morning (n = 4) or evening (n = 2) during RH coadministration [17, 35, 38, 49 65], while one study only reported body weight-specific EFVC 12 data during RH coadministration [66] and one reported only proportions of patients with subtherapeutic EFV concentrations [67]. General characteristics of studies Table 1 shows characteristics of the 22 included studies. All were published between 2006 and 2016 with 14 (64%) and seven (32%) conducted exclusively in Africa and Asia, respectively. The majority of studies included adult patients (91%). The two child studies were conducted in South Africa. In adults, EFV was systematically administered at a dose of 600 mg day 1 of which one study also had EFV 800 mg day 1 dose administered to a selected comparative group of patients [35]. EFV was administered at bed-time to improve the tolerability and reduce adverse events [68], except for four adult studies where it was taken in the morning [17, 51, 53, 61] and a 6-month anti-tb treatment was used and administered daily at recommended dosing in all studies except one with intermittent administration [53]. Br J Clin Pharmacol (2018)

4 D. Atwine et al. Figure 1 Flow chart showing the selection of relevant papers Thirteen studies (59%) reported EFV concentrations on and off RH within the same patients. All studies had EFV concentrations during RH coadministration, with sampling done at steady state during intensive phase or continuation phase of TB treatment. Individual study-specific characteristics are shown in Appendix S2. Effect of RH coadministration on the PK of EFV in HIV/TB coinfected patients Overall, 19/20 studies reported the median or mean EFV C 12 or C min within the allegated therapeutic window ( ng ml 1 ) during RH coadministration irrespectiveofthegeographicalregion(table2).onestudyinthai patients [50] reported median or mean EFV C 12 or C min > 4000 ng ml 1 although without specifying the proportion of patients with supratherapeutic concentrations. Of the 14 studies conducted among adult patients that reported EFV concentrations on and off RH, 10 were in Africa and four in Asia. Notably, among the 10 adult studies conducted in Africa, six reported higher EFV concentration during RH coadministration compared to the off-rh period, although the difference is highly variable, ranging from 3.7% to 33.3% across studies and countries [17, 35, 51, 60, 62, 65]. The remaining four adult African studies reported a lower EFV concentration while on RH, still with a difference that is highly variable, ranging from 16.3 to 33.3% [54 56, 64]. Higher EFV concentrations during RH coadministration was also observed in two adult studies conducted in South-East Asia [50, 53], with the remaining two studies [57, 63] reporting reduced EFV concentrations. Of the two studies among the South African children, the earlier study by Ren et al. [52]indicatedunchangedmedian 1644 Br J Clin Pharmacol (2018)

5 Pharmacokinetics of efavirenz with antituberculosis treatment Table 1 Characteristics of the selected studies and of the human immunodeficiency virus (HIV) tuberculosis (TB) coinfected patients included Characteristic Number of studies Study types, n (%) 22 Prospective nonrandomized comparative study 6 (27.3) Cohort from randomized clinical trial 8 (36.4) Prospective cohort 6 (27.3) Case control 1 (4.6) Cross-sectional 1 (4.6) Geographical setting of studies, n (%) 22 Africa 14 (63.6) Asia 7 (31.8) Africa/Latin America 1 (4.6) Study population, n (%) 22 Adult 20 (90.9) Children 2 (9.1) Age in years, median [range] Adult studies [ ] Child studies [ ] Sex, male Adults, n (%) 20 <50% 5 (25.0) 50 70% 11 (55.0) >70% 4 (20.0) Children, n (%) <50% 2 1 (50.0) 50 70% 1 (50.0) Body weight (kg) Adult studies, median [range] [ ] Child studies, median [range] [ ] Sample size for selected studies, median [IQR] [45 270] Adults, n (%) 20 <30 3 (15.0) (25.0) > (60.0) Children, n (%) 2 <30 1 (50.0) (50.0) Patient groups, n (%) 22 HIV-TB coinfected and HIV-mono infected (2 parallel groups) 7 (31.8) HIV-TB coinfected on and off anti-tb drugs 13 (59.1) HIV-TB coinfected on anti-tb drugs (no off anti-tb drugs period) 2 (9.1) (continues) Br J Clin Pharmacol (2018)

6 D. Atwine et al. Table 1 (Continued) Characteristic Number of studies Anti-TB treatment administration frequency for 6 months, n (%) 22 Daily 21 (95.5) Three times weekly 1 (4.5) EFV intake, n (%) 22 Evening 20 (90.9) Morning 2 (9.1) EFV Pharmacokinetic parameter reported*, n (%) 22 C12 (mid-dose concentration) 16 (72.7) Cmin (Ctrough, predose concentration) 6 (27.3) CNStoxicityassessed,n (%) 22 8 (36.4) Hepatotoxicity assessed, n (%) 22 7 (31.8) Virological response assessed following standard anti-tb drugs coadministration, n (%) (54.5) EFV, efavirenz; C12, mid-dose concentration; Cmin, Ctrough, predose concentration; CNS, central nervous system; IQR, interquartile range C 12 with RH coadministration (0.8%), while a recent study by McIlleron et al. [38] indicated a decrease of 16.3%. Subtherapeutic EFV concentration and virological response Figure 2 shows the 13 studies reporting the proportion of patients with subtherapeutic EFV levels on or off RH. The proportion of patients with subtherapeutic EFV concentration (<1000 ng ml 1 ) on RH range between 3.1 to 72.2% in 12 studies including one conducted in children. Ten studies, had proportions for both on and off RH, with nine reporting higher proportions of patients with EFV < 1000 ng ml 1 during RH coadministration than without RH, although the difference was highly variable, ranging between 1.1 and 21.1%. Only two studies, one in Thai patients [57] and one in South Africans [61], reported a lower proportion of patients with EFV < 1000 ng ml 1 during RH coadministration than without RH, with a difference <5%. Two studies (one in adult and one in children) reported a very high proportion of patients (>50%) with subtherapeutic concentrations either with or without RH [52, 67]. These striking results could possibly be due to an adherence issue, genetic polymorphism or low EFV dosing in the child study. Table 3 shows results of virological response with EFV and RH coadministration. Out of 10 studies (eight in adults, two in children) reporting both EFV exposure and virological suppression, six adult studies reported a proportion of patients with subtherapeutic EFV concentrations ranging from 3 to 32% although the virological suppression was 80% between 6 and 12 months follow-up. Of them, four studies had <20% of patients with subtherapeutic concentrations. In contrast, both studies (all in adults) with virological suppression <80% had >20% [range: 27 72%] of patients with subtherapeutic levels [59, 67]. Although both child studies reported a high virological suppression (~ 85%). In the study by Ren et al. [52] there was an obvious miss-match between the reported proportion of children with subtherapeutic EFV Cmin (60%) and the reported high virological suppression (84.6%). Supratherapeutic EFV concentration and safety Figure 3 shows the seven studies that reported a proportion of patients having supratherapeutic EFV concentrations (>4000 ng ml 1 ) on RH. Five studies had data on proportion of patients with EFV > 4000 ng ml 1 during and without RH coadministration. Five studies [17, 59, 61, 63, 65], all in adults, reported higher proportions of patients with EFV > 4000 ng ml 1 during RH as compared to off RH, although the difference was highly variable, ranging between 3.0 and 23.1%. Table 4 shows the hepatic adverse events among HIV-TB coinfected patients. Of the eight studies that assessed and reported data on hepatic events, the four African adult studies reported incidence of any grade alanine aminotransferase (ALT) rises ranging from 2.8 to 30% among TB-HIV coinfected patients [35, 51, 60, 61], although a relationship between ALT rise and EFV/RH coadministration was demonstrated in only one study [60]. Among the three Asian studies [58, 63, 67], incidence of any grade ALT rises ranged between %, with a significant relationship, with EFV/RH coadministration reported only in one study [63]. The only child study conducted in South Africa reported all grade ALT rises in 2.5% of children [38]. Table 4 shows the CNS adverse events among HIV-TB coinfected patients. Nine studies (six African, two Asian and one international) assessed and reported data on incidence of any grade CNS symptoms. Among African studies, two showed a significant relationship between CNS adverse events and supratherapeutic EFV plasma concentrations [55, 1646 Br J Clin Pharmacol (2018)

7 Pharmacokinetics of efavirenz with antituberculosis treatment Table 2 Concentrations of efavirenz (EFV) coadministered with rifampicin isoniazid (RH) and without RH or after the discontinuation of tuberculosis (TB) treatment, measured at steady state ( 4 weeks) in 20 selected studies. Concentrations (ng ml 1 ) are reported as mean (standard deviation), median (interquartile range) or median [range] Region/study description Author, year (Country) Ref. Comparison group On RH Off RH % Difference in EFV concentration between on n Cmin C12 n Cmin C12 and off-rh Africa Five adult studies, with on/off RH data in same patient Semvua, 2013; (Tanzania) Bhatt, 2015; (Mozambique) Bienvenu, 2014;(Rwanda) [46] S ( ) ( ) 8.3% [65] S ( ) ( ) 3.7% [64] S ( ) ( ) 33.3% Friedland, 2006; [17] S [ ] [ ] 25.4% (South Africa) Five adult studies, with on-rh data in TB-HIV coinfected group and off-rh data in HIV only patient group Two child studies with on/off RH data in same patient Two studies in adults, without off RH data Orrell, 2011; (South Africa) Ngaimisi, 2011; (Tanzania) Habtewold, 2015; (Ethiopia) Cohen, 2009; (South Africa) Mukonzo, 2014; (Uganda) Mukonzo, 2013; (Uganda) Ren, 2009; (South Africa) McIlleron, 2013; (South Africa) Yimer, 2011; (Ethiopia) Gengiah, 2015; (South Africa) [35] S ( ) ( ) 9.1% [54] P ( ) ( ) 28.9% [60] P ( ) ( ) 17.4% [62] P ( ) ( ) 33.3% [56] P ( ) ( ) 17.1% [55] P ( ) ( ) 24.5% [52] S ( ) 1240 ( ) ( ) 1230 ( ) 0.8% [38] S ( ) ( ) 16.3 [49] S ( ) [61] S ( ) Asia Three adult studies, with on/off-rh data in same patient Ramachandran, 2013 c ;(India) Uttayamakul, 2010; (Thailand) [53] S (2500) (1900) 9.5% [50] S (5970) (2670) 26.3% (continues) Br J Clin Pharmacol (2018)

8 D. Atwine et al. Table 2 (Continued) Region/study description Author, year (Country) Ref. Comparison group On RH Off RH % Difference in EFV concentration between on n Cmin C12 n Cmin C12 and off-rh Borand, 2014; (Cambodia) [63] S (1753,4494) ( ) 3.6% One adult study, with on-rh data in TB-HIV coinfected group and off-rh data in HIV only patient group Manosuthi, 2013; (Thailand) [57] P ( ) ( ) 22.2% One adult study without off-rh data Manosuthi, 2009; (Thailand) [58] None (3780) Africa/Latin America One study without off-rh data Luetkemeyer, 2013; (International) [59] S ( ) Geometric mean (90% confidence intervals) or geometric mean [range]; C12, mid-dose concentration; Cmin, Ctrough, predose concentration; Ref.no, reference number corresponding to the cited study. S, pharmacokinetic comparisons done in same patients; P, pharmacokinetic comparisons done in different patients; This was calculated as the difference in mean or median C12 or Cmin during on RH and off RH, as a fraction of the mean or median C12 or Cmin during on RH, and expressed as a percentage. This estimates the change in EFV plasma concentration attributable to RH coadministration HIV, human immunodeficiency virus 1648 Br J Clin Pharmacol (2018)

9 Pharmacokinetics of efavirenz with antituberculosis treatment Figure 2 Proportion of patients with subtherapeutic efavirenz (EFV) plasma concentration. Child studies; S, on and off rifampicin isoniazid (RH) EFV subtherapeutic concentrations assessed in the same patient; P, on and off RH EFV subtherapeutic concentrations assessed in different patient groups (HIV-TB coinfected vs.hivonly).n, sample size on which the EFV subtherapeutic frequencies either with or without RH coadministration is based; n1, sample size on which the EFV subtherapeutic frequencies during ART without RH coadministration is based; n2, sample size on which the EFV subtherapeutic frequencies during ART with RH coadministration is based; ref = reference number 62]. Of the two Asian studies, one reported a significant relationship between developing incidence of CNS side-effects of any grade with having at least one-time supratherapeutic EFV concentrations [63]. EFV concentration by body weight, EFV dose and sex Three studies reported plasma EFV concentrations stratified by body weight when given at a 600 mg dose with anti-tb drugs coadministration [59, 60, 66]. Two studies reported lower EFV concentrations in patients with weight > 50 kg than in patients with weight < 50 kg, with median C 12 (interquartile range, IQR) of 2060 ng ml 1 ( ) vs ng ml 1 ( ) in Cambodian patients [66] and mean C min of 1860 ng ml 1 vs ng ml 1 in a study that included African, Latin American and Asian sites [59]. By contrast, in the study conducted in Ethiopia, the median (IQR) C 12 was slightly higher in patients with weight > 50 kg compared to those with weight < 50 kg: 1515 ng ml 1 ( ) vs ng ml 1 ( ) [60]. However, in the same study, without coadministration, there was a trend towards lower concentrations in patients with higher body weight: 1233 ng ml 1 [848, 1670] vs ng ml 1 [1067, 2155] [60]. Br J Clin Pharmacol (2018)

10 D. Atwine et al. Table 3 Presentation of virological response with efavirenz concentrations during rifampicin isoniazid coadministration Studies Reference number VL threshold (copies ml 1 ) Follow-up (months) % patients with Subtherapeutic levels * %patients with VL suppression Adult studies, n =8 Mariana, 2016 (Indonesia) [67] <40 3to Manosuthi, 2009 (Thailand) [58] < Habtewold Abiy, 2015 (Ethiopia) [60] < Bhatt, 2015 (Mozambique) [65] < Friedland, (South Africa) [17] < Borand, 2014 (Cambodia) [63] < Luetkemeyer, 2013 (Botswana, Brazil, Haiti, Kenya, Malawi, South Africa, Thailand, Uganda, Zimbabwe, Peru, USA) [59] < * 71.4 Orrell, 2011 (South Africa) [35] < Child studies, n =2 Ren, 2009 (South Africa) [52] < * 84.6 McIlleron, 2013 (South Africa) [38] < All results based on C12 unless otherwise indicated *in case of Cmin based results; VL, viral load Only one study in South Africa included patients on high EFV dose (800 mg) [35]. A higher median C 12 of EFV during RH coadministration was noted in the patients on 800 mg EFV (2900 ng ml 1,IQR: ) as compared to those on600mgefv(2400ngml 1,IQR: ). Only one study presented plasma EFV concentrations stratified by sex during coadministration with anti-tb drugs. The mean C min was lower for males than for females (1870 vs ng ml 1 ), which could be related to differences in patients body weight [59]. The frequency of CYP2B6 slow and extensive metabolizers, and effect on EFV concentration during anti-tb treatment coadministration A total of nine studies (six from Africa and three from Asia) reported EFV concentrations according to the CYP2B6 G516T genetic polymorphism encoding for a defective enzyme, and eight of nine studies (five from Africa and three from Asia) reported the frequency of CYP2B6 homozygous slow metabolizer genetic polymorphism within the studies populations. Most of the studies reported only frequencies for the most frequent polymorphism CYP2B6 516 G > T (CYP2B6*6 allele). In all studies, except for one conducted in Tanzania [54], patients who carried this loss of function allele had higher EFV concentrations both off and on RH as compared to those carrying the wild-type gene as shown in Figure 4, panels A and B respectively. The frequency of slow metabolizers ranged between 10% in one study conducted in Rwanda [64] and 28% in another study in Ethiopia [60], while the frequency of extensive metabolizers ranged from 34 to 50% across studies. Discussion In this review, we note that all selected studies, apart from one conducted in children [52], reported median or mean EFV C 12 or C min within the allegated therapeutic range ( ng ml 1 )duringrh-basedstandardtbdrugcoadministration, hence supporting the recommended 600 mg EFV dosing in African or Asian HIV/TB coinfected patients. In addition, many of these studies also showed an increase in EFV concentrations during RH coadministration. Notably, the two studies that enrolled >200 patients [63, 65] reported a very small difference (~4%) in median EFV C 12 with vs. without RH coadministration. Surprisingly, both studies reported some patients having higher EFV concentrations on vs. off RH. This observation was demonstrated to be dependent on CYP2B6 and NAT2 genetic polymorphism. Indeed, patients who are CYP2B6 slow metabolizers, had higher concentrations of EFV (>4000 ng ml 1 ). In those patients, R has little effect on minor drug metabolizing enzymes involved in EFV biotransformation, although H, which is metabolized by the polymorphic NAT2, was demonstrated to inhibit these enzymes [40, 69] leading to higher EFV concentrations on RH vs. offrhasshowninfigure4a. In contrast, extensive metabolizers have lower EFV concentrations with little or no effect of RH as shown in Figure 4B. In summary, this drug drug interaction is complex and owing to the difference in frequencies of genetic polymorphism of CYP2B6 and NAT2 enzymes, EFV concentrations may be higher or lower when coadministered with RH or administered alone [33, 70]. Studies conducted among adult patients in Tanzania and Thailand or in children in South Africa highlight within 1650 Br J Clin Pharmacol (2018)

11 Pharmacokinetics of efavirenz with antituberculosis treatment Figure 3 Proportion of patients with supratherapeutic efavirenz plasma concentration. Child studies; EFV, efavirenz; RH, rifampicin and isoniazid; S, on and off RH subtherapeutic EFV concentrations assessed in the same patient; P, on and off RH subtherapeutic EFV concentrations assessed in different patient groups (HIV-TB coinfected vs. HIVonly).n, Sample size on which the EFV subtherapeutic frequencies either with or without RH coadministration is based; n1, Sample size on which the EFV subtherapeutic frequencies during ART without RH coadministration is based; n2, Sample size on which the EFV subtherapeutic frequencies during ART with RH coadministration is based; ref, reference number country variability with regard to effect of RH on EFV concentrations [38, 50 52, 54, 56, 57], which could partially be explained by the small sample size, or differences in methodology used and the pharmacogenetics within different ethnic groups. As highlighted in the review by Colic et al. [71], the reported interpopulation variability in EFV exposure among African and Asian countries could be due to the higher genetic diversity with regard to CYP2B6 among individuals in different population groups or ancestral origins [72]. The influence of age on CYP2B6 expression has not been well established although previous studies hypothesized that it may also depend on sex, as significant increase in liver CYP2B6 is more linked to only males at higher age [73]. The lower EFV concentrations among males reported in one study [59] are in agreement with what has been reported in another study in Zimbabwe without anti-tb drugs coadministration that showed a mean EFV C 12 lower in males than in females [74] although this study was excluded in this review given a non-specified timing of PK sampling. This might also be dependent on the CYP2B6 genetic polymorphism and age [73]. With the few studies available, it was not possible to satisfactorily assess the effect of body weight and EFV dose during coadministration with anti-tb drugs. Nevertheless, the effect of body weight if any is small and does not warrant dose optimization. Similarly, no strong conclusions could be made with regard to the EFV exposure during RH coadministration in children, given that only two studies are available [38, 52]. The observed differences in the effect of RH on EFV in these two studies conducted in South Africa could not be explained by age of the children but could have been driven, first by the differences in sample sizes with one study performed among 15 children [52] and the other among 40 children [38]. Br J Clin Pharmacol (2018)

12 D. Atwine et al. Table 4 Hepatic and central nervous system (CNS) adverse events among human immunodeficiency virus (HIV) tuberculosis (TB) coinfected patients Study Ref no Number of HIV-TB coinfected patients included in the analysis, n %ofpatients with all grade ALT increase %ofpatients with all grade CNS events Relationship between ALT rise with EFV/R coadministration Relationship between CNS events with EFV/ R coadministration Mariana, 2016; (Indonesia): P Habtewold, 2015; (Ethiopia): P Borand, 2014; (Cambodia): S [67] No relationship despite the higher incidence of all-grade ALT rise among HIV-TB patients (16.7%) vs. HIV alone (3.7%). No patient developed supratherapeutic EFV concentrations. [60] No data Relationship with EFV/R coadministration noted. Incidence of grade 3ALTrise higher in HIV-TB patients on EFV/ R coadministration (30%) than in HIV patients on EFV alone (15.7%). The role of high EFV plasma concentration and CYP2B6*6 genotype noted in both patient groups. NAT2 slowacetylator genotype, as determined by sequencing of NAT-2 coding region predicted liver toxicity in TB-HIV coinfected on isoniazid. Overlapping drug toxicity (ART and anti-tb drugs) and disease effect (TB-HIV coinfection) could not be ruledout. [63] No relationship noted with grade 3 transaminase elevation (P =0.30),insteadasignificant relationship was noted between the risk of any grade hepatotoxicity with having consistent supratherapeutic EFV concentrations [P < 0.001, OR=1.52( )] but not with intermittent EFV levels >4000 ng ml 1,ascompared with those in normal ranges. No relationship noted. Since no patient had supratherapeutic EFV plasma concentrations, Authors attributed the high onset of CNS events to EFV s high lipophillic nature allowing it to easily penetrate the blood brain barrier. Not applicable No relationship noted with CNS events grade 3, P = 0.30, but a significant relationship was noted between developing a CNS side-effect irrespective of grade with having at least one time supratherapeutic EFV levels as compared to those with normal ranges, P < 0.001, OR = 2.72 [ ] Mukonzo, 2013; (Uganda): P Luetkemeyer, 2013; (International): S [55] 138 No data 74.0 Not applicable Relationship with EFV/R coadministration noted. All grade CNS symptoms during ART were significantly predicted by EFV plasma concentrations consistently. No significant differences in incidence of CNS symptoms betweenpatientsonefvwith R (74%) and those without R (72%) cotreatment (P =0.73) was noted. No treatment discontinuation occurred due to severe CNS events. [59] 780 No data 5.9 No relationship with EFV/R coadministration noted. EFV Cmin >4000 ng ml 1 was not significantly associated with occurrence of grade 3 or higher CNS events. (continues) 1652 Br J Clin Pharmacol (2018)

13 Pharmacokinetics of efavirenz with antituberculosis treatment Table 4 (Continued) Study Ref no Number of HIV-TB coinfected patients included in the analysis, n %ofpatients with all grade ALT increase %ofpatients with all grade CNS events Relationship between ALT rise with EFV/R coadministration Relationship between CNS events with EFV/ R coadministration Friedland, 2006; (South Africa): S McIlleron, 2013; (South Africa): S Bhatt, 2015; (Mozambique): S Orrell, 2011; (South Africa): S Cohen, 2009; (South Africa): P [17] 19 No data 36.8 No clear association was observed between onset of all grade CNS symptoms and plasma EFV levels. [38] No relationship with EFV/R coadministration noted. Only 1 child suffered a grade 3 elevation in ALT in the month after completion of anti-tb treatment, which turned to normal without treatment adjustment. This child had an average mid-dose interval concentration of 17.7 mg l 1 during anti-tb treatment, which dropped to 4.14 mg ml 1 a month after stopping R and isoniazid. There was a low incidence of liver toxicity with use of R 10 mg kg 1. No relationship with EFV/R coadministration noted. Though assessed, no grade 3 or 4 CNS events recorded. Subtle effects were not recorded in the study. Lack of CNS events was linked to good tolerability given the night administration of EFV. [65] 302 No data 2.0 Not applicable No relationship with EFV/R coadministration noted. No significant association between occurrence of grade 2orhigherCNSadverse events reported within the first 12 weeks of ART and EFV concentrations >4000 ng ml at week 12, P =0.293 [35] No data No relationship with EFV/R coadministration noted. Both grade 2 and 3 events of raise in ALT occurred during EFV coadministration. The absence of ALT elevation events before ART initiation signified that these events were EFV-related. No direct link made to EFV interaction. Not applicable [62] 137 No data 35.8 Not applicable A relationship with EFV/R coadministration noted. About 31% of those with CNS symptoms (all grade) had high EFZ concentrations. No significant associations between EFV concentrations and other neuropsychiatric symptoms. Semvua, 2013; (Tanzania): S Gengiah, 2015; (South Africa): S [51] No data No relationship with EFV/R coadministration noted. Only ALT rises of grade 1 noted, with no link with EFV interaction established. [61] No relationship with EFV/R coadministration noted. Approximately 67% of all events of transaminase rise (grade 3 or 4) occurred during antitubercular treatment alone, Not applicable No relationship with EFV/R coadministration noted. All the reported 3 CNS toxicity events (all grade), were from 1 patient on EFV 800 mg and hadacmin=2100ngml 1 at (continues) Br J Clin Pharmacol (2018)

14 D. Atwine et al. Table 4 (Continued) Study Ref no Number of HIV-TB coinfected patients included in the analysis, n %ofpatients with all grade ALT increase %ofpatients with all grade CNS events Relationship between ALT rise with EFV/R coadministration Relationship between CNS events with EFV/ R coadministration and only 26.7% during antitubercular treatment/art. No link to interaction and events during ART resolved without drug cessation. the time of onset of CNS events. Symptoms ceased with a switch from morning to night administration of EFV. Manosuthi, 2009; (Thailand): S [58] 71 0 No data No relationship with EFV/R coadministration noted. No NNRTI-associated hepatitis with EFV coadministered with R. Not applicable S, same patient comparisons (only HIV-TB coinfected). P, Parallel patients comparisons (both HIV only and HIV-TB coinfected) ALT, alkaline aminotransferase; EFV, efavirenz; R, rifampicin; ART, antiretroviral therapy; Cmin, minimum EFV concentration; Ref no, reference number Second, the dose difference of EFV used in the two studies. Third, the frequency of genetic polymorphism, given the genotype frequencies in different ethnic African populations as previously reported [71, 75]. Although the proportion of patients with subtherapeutic EFV levels was very high in some studies, there were no major differences with and without anti-tb treatment. The frequency of CYP2B6 extensive metabolizers, could be a plausible explanation for these subtherapeutic EFV concentrations, although lack of adherence can not be ruled out [48, 76]. As expected, there was a trend of lower virological response in studies with very high proportion of patients with subtherapeutic concentrations. However, some discordances between the EFV concentrations and the virological response observed in some studies [17, 52, 60], illustrate the difficulty to correlate the drug plasma concentration measured at one point of time with the virological response. This is in agreement with the PK/pharmacodynamic results of the ENCORE1 study, where patients were randomized to receive EFV once daily either at 400 mg or 600 mg. It was shown that despite reported C 12 < 1000 ng ml 1 in 5% and 2% for EFV400 and EFV600 respectively, one patient in the EFV400 group and three in the EFV600 group had detectable plasma viral load at 48 weeks of therapy [77]. In addition, it highlights the limitation of the commonly used subtherapeutic threshold of 1000 ng ml 1 for mid-dose EFV concentration, which is based on very low level of evidence [18]. Furthermore, the use of the same threshold for studies reporting C min (trough concentration) concentrations could have led to an overestimation of the proportion of patients with subtherapeutic concentrations [52]. The occurrence of supratherapeutic EFV levels was very common both in African and Asian studies. Higher EFV concentrations during coadministration with anti-tb treatment could increase the occurrence of adverse events [11, 19, 20]. EFV CNS adverse effects have been reported to be more common in those patients with higher EFV concentrations [18, 48, 78]. However, due to the low number of studies reporting both safety and PK data during coadministration with anti-tb treatment, it is difficult to draw strong conclusions based on this current review. In addition, the lack of information or standardisation in reporting safety information between studies, especially for CNS adverse events, makes the interpretation even more difficult. Nevertheless, we note that no clear correlation could be made between EFV supratherapeutic levels and occurrence of CNS adverse events during RH coadministration within the exclusively African studies [17, 55, 65]. This lack of association might also be biased by the other common causes of neuropsychiatric disorders besides EFV treatment in HIV infected patients [79, 80]. Some studies have, however, attempted to explain this disparity between plasma EFV concentrations and onset of CNS adverse events on grounds of the high lipophilic nature of EFV which allows it to penetrate the blood brain barrier easily and so give disproportionate EFV concentrations between plasma and brain [67]. Interestingly, it was recently suggested that among 563 patients who had been initiated on EFV-containing regimens at an HIV primary care clinic in the south-eastern USA, slow metabolizer genotypes were significantly associated with EFV discontinuation due to onset of CNS symptoms, although this association was considerably stronger in Whites than in Blacks [81]. Regarding, hepatotoxicity, the reported overall incidence of ALT rises to any grade among TB-HIV coinfected patients was higher among the African adult studies [2.8 30%] as compared to Asian studies [0 16.7%]. Nevertheless, only one adult study indicated a significant relationship between any grade ALT rise with EFV and RH coadministration [63]. In this review, it was not possible to distinguish the individual drug contribution of R, H or EFV on ALT rises of any grade given that the timing of onset was not well clarified in all studies. In one child study conducted in South Africa, all grade ALT rises were noted in 2.5% of children, with only one child suffering a grade 3 elevation in ALT in the month after completion of anti-tb treatment, which turned to normal without treatment adjustment [38]. This systematic review has some limitations: (i) the great heterogeneity between studies with regard to study designs, PK parameters explored, and reporting, hindered any 1654 Br J Clin Pharmacol (2018)

15 Pharmacokinetics of efavirenz with antituberculosis treatment Figure 4 Efavirenz plasma concentrations on and off rifampicin coadministration among patients with CYP2B6 homozygous slow and extensive metabolizer genetic polymorphism, CYP2B6 G516 T. (A) Slow metabolizer genetic polymorphism; (B) extensive metabolizer genetic polymorphism. On-R, on rifampicin; Off-R, off rifampicin; SM, slow metabolizer; EM, extensive metabolizer; EFV, efavirenz; Ref, reference number; C12, efavirenz mid-dose concentrations; Cmin, minimum efavirenz concentration; A, C12; mean; B, C12; median; C, Cmin; median; D, Cmin; mean potential meta-analysis; (ii) the small sample sizes for TB-HIV coinfected populations in many studies, may have contributed to the observed variability in EFV exposure due to RH coadministration even within same country; (iii) most studies did not attempt to correlate subtherapeutic and supratherapeutic concentrations of EFV during RH coadministration with CYP2B6 genetic polymorphisms, and so hindered a clear explanation of the observed changes; (iv) the few studies which enrolled children could not allow a thorough evaluation and conclusions on EFV exposure with RH coadministration this needs to be highlighted as children are a vulnerable population who need optimized dosing for improved efficacy; (v) analysis of safety information was limited by the low number of studies correlating both safety, body weight and sex and PK data and by the variability in theassessmentofcnsadverseevents,withonlyonestudy using a standard scale [55]. Conclusion This systematic review shows a minimal effect of RH coadministration on EFV plasma concentrations, when EFV is used at a 600 mg dosing. This supports the current recommendation for coadministration of ART regimen with EFV 600 mg daily and anti-tb treatment in TB-HIV high burden countries. The CYP2B6 genetic polymorphism is the more likely explanation for the variability of EFV concentrations in African and Asian patients on coadministration with anti-tb treatment. The interpretation and management of elevations in ALT and CNS adverse events should be done not only in the context of EFV and RH interaction but also looking at other independent predictors such as advanced disease, liver diseases, adherence and patient characteristics. This systematic review is important as ritonavir/cobicistat boosted PI cannot be used with R, and sufficient data are not yet available on potential use of raltegravir or dolutegravir [82, 83]. Initial PK data for coadministration of EFV at 400 mg dose with RH were recently reported showing no major reduction in EFV concentration suggesting that EFV 400 mg plus RH could be safe. However, the full report of the results is yet to be published [84]. Since new TB drugs allowing shorter TB treatment regimen will not be available soon, there is need to optimize the current first line drugs for susceptible TB. Increasing the rifampicin dose is an option [85] raising the issue of the drug drug interaction with EFV. The ANRS Rifavirenz trial has recently shown a minimal effect of rifampicin at 20 mg kg 1 dosing on EFV exposure [86]. Because of the discrepancies between the two child studies, there is a need for better evidence to guide on the EFV dosing during anti-tb drug coadministration in this population. Competing Interests There are no competing interests to declare. Br J Clin Pharmacol (2018)